• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.6
• /
• pp.319-325
• /
• 2023

In this study, we proposed a method for hyperparameter optimization in the building and training of a deep learning model designed to process point cloud data collected by a millimeter-wave radar system. The primary aim of this study is to facilitate the deployment of a baseline model in resource-constrained IoT devices. We evaluated a RadHAR baseline deep learning model trained on a public dataset composed of point clouds representing five distinct human activities. Additionally, we introduced a coarse-to-fine hyperparameter optimization procedure, showing substantial potential to enhance model efficiency without compromising predictive performance. Experimental results show the feasibility of significantly reducing model size without adversely impacting performance. Specifically, the optimized model demonstrated a 3.3% improvement in classification accuracy despite a 16.8% reduction in number of parameters compared th the baseline model. In conclusion, this research offers valuable insights for the development of deep learning models for resource-constrained IoT devices, underscoring the potential of hyperparameter optimization and model size reduction strategies. This work contributes to enhancing the practicality and usability of deep learning models in real-world environments, where high levels of accuracy and efficiency in data processing and classification tasks are required.

• The Journal of the Korea Contents Association
• /
• v.9 no.4
• /
• pp.189-197
• /
• 2009

A digital image usually has 8 bits of depth basically representing pixel intensity ranging for [0 255]. These pixel range allow 256 step levels of pixel values in the image. Thus the greyscale value for a given image is an integer. When we carry out interpolation of a given image for resizing we have to round the interpolated value to integer which can result in loss of quality on perceived color values. This paper proposes a new method for recovering this loss of information during interpolation process. By using the proposed method the pixels tend to regain more original values which yields better looking images on resizing.

• The Journal of the Korea Contents Association
• /
• v.12 no.8
• /
• pp.87-95
• /
• 2012

We propose that a subjective image quality assessment based on objective image quality factors in order to evaluate objectively preference of consumers. In other words, we define objective image quality factors which are easy to accept by manufacturers and they are composed of subjective image quality assessment questionnaires. Also, portrait image is selected by stimulus in order to persue easiness of evaluation for the general subjects. Throughout a subjective image quality assessment model, we evaluate recognition of image quality by consumers and analyze the effectiveness of correlation in terms of the final image quality preference. Analyzing the relationship between image quality factors, we can figure out the preferable image quality and confirm the positive effects on consumers' recognition of image quality. In the results, there are strong relationship between preference and color reproduction, dynamic range, noise, and resolution respectively. especially, the characteristic of portrait, there is high correlation between color reproduction and preference.

• Journal of Korea Multimedia Society
• /
• v.15 no.2
• /
• pp.179-194
• /
• 2012

Digital image forgery is becoming a topic of great interest with regard to honesty in imaging. We can often see forged digital images in a variety of places, such as the internet, and magazines, and in images used in political ads, etc. These can reduce the reliability and factual basis of the information contained in image. Therefore, objectivity is needed to determine if the image is forged so as to prevent confusion in the viewing public. Most digital forgeries consist of image resizing, rotating including the following interpolations. To find evidence of interpolation in forged images, this paper proposes a new method for detecting digital image forgery using general interpolation factors analyzed through re-interpolation algorithm of the forged images in order to determine the differences in the patterns. Through the re-interpolation algorithm we could detect the forged region and filtering region used image retouching included to interpolation.

• Asian Pacific Journal of Cancer Prevention
• /
• v.13 no.9
• /
• pp.4623-4626
• /
• 2012

Background: The risk of treatment-related death (TRD) and febrile neutropaenia (FN) with adjuvant taxane-based chemotherapy for early breast cancer is unknown in Malaysia despite its widespread usage in recent years. This study aims to determine these rates in patients treated in University Malaya Medical Centre (UMMC). Patients and Methods: Patients who were treated with adjuvant taxane-based chemotherapy for early breast cancer stages I, II or III from 2007-2011 in UMMC were identified from our UMMC Breast Cancer Registry. The TRD and FN rates were then determined retrospectively from medical records. TRD was defined as death occurring during or within 30 days of completing chemotherapy as a consequence of the chemotherapy treatment. FN was defined as an oral temperature > $38.5^{\circ}C$ or two consecutive readings of > $38.0^{\circ}C$ for 2 hours and an absolute neutrophil count < $0.5{\times}10^9/L$, or expected to fall below $0.5{\times}10^9/L$. Results: A total of 622 patients received adjuvant chemotherapy during this period. Of these patients 209 (33.6%) received taxane-based chemotherapy. 4 taxane-based regimens were used namely the FEC-D, TC, TAC and AC-PCX regimens. The commonest regimen employed was the FEC-D regimen accounting for 79.9% of the patients. The FN rate was 10% and there was no TRD. Conclusion: Adjuvant taxane-based chemotherapy in UMMC for early breast cancer has a FN rate of 10%. Primary prophylactic G-CSF should be considered for patients with any additional risk factor for FN.

• Korean Journal of Animal Reproduction
• /
• v.19 no.1
• /
• pp.55-63
• /
• 1995

The putative totipotency germ cells has a relative abundance of alkaline phosphatases. Thus, histological staining of AP activity offers a new route to isolate totipotent cells and also provides insights into culture systems of these cells. Furthermore, the AP staining technique is simple and fast, requires only the napthol AS/MS substrate in combination with trapping diazonium salts such as fast red or fast blue. However, our unexpected finding was that AP staining of mouse ES cells were detected in the undifferentiaed epiblast-derived cells as well as several types of differentiating cells. This findings are different from results of Talbot et al. (1993) reported usefulness of the AP staining and implies that histological staining of AP may not by useful to determine undifferentiaed state or totipotency of ES cells. Thus, we have investigated the patterns of AP expression by RT-PCR in order to identify a marker of undifferentiated ES/primordial germ (PG) cells. In RT-PCR analysis, embryonic (E)-AP was detected only in undifferentiated ES cells, but intestinal(I)-AP was not detected in all of the examined ES and PG cells. In addition, nonspecific (NS)-AP wasdetected in undifferentiated PG cell from day 7, 5 to 13 of gestation. Histological activity of AP in ES cells was completely suppressed by addition of L-phenylalanine (Phe), L-homoarginine (Har), and L-phenylalanylglycylglycine (PheGlyGly) as an inhibitor, but RT-PCR showed the same results as in the absence of an inhibitors. Our findings suggested that expression of E-AP and NS-AP may use as a marker to determine the undifferentiated status in ES and PG cells.

• Journal of Nutrition and Health
• /
• v.11 no.1
• /
• pp.1-8
• /
• 1978

Ninety-three weaning children, 8-36 Mo. old wererandomly selected from rural areas of South Korea. Food and nutrient intake level were measured by the weighing-interview method for two consecutive days in December 1976. Their weaning pattern, dietary pattern, food and nutrient intake level were analyzed. A brief summary of 4he finding follows; 1) 80% of the rural children completed the weaning by the age of 24 months. 2) 44% of 8-12 month olds were living only on breast mil while the rest of the age group received regularly a supplementary food. 3) Over half of the 8-36 Mo old children investigated were receiving the same dietary pattern, arbitrarily called 'Diet K'. Diet K consists of rice-Kimchi-seasoning-oil which is the basic components of an average Korean adult diet. This very composition of the diet K and no other variety of food included inevitably result in low intake of vitamin A, iron, calcium and riboflavin for the children. 4) The weaning children were found to be taking all nutrients except niacin in amount far from sufficient to recommended level. Particularly low were the intakes of good quality protein, iron, vitamin A, riboflavin and vitamin C which met only 15%-37% of the recommended level for 12-36 Mo. olds. 5) It was concluded that majority of the rural children of weaning age in Korea are given the simple diet K and feeding the weaning children only with the basic dietary pattern of adults can not meet the nutritional requirements. This is a part of the study supported by the ROKG/USAID Health Planning Project.

• The Journal of the Korea Contents Association
• /
• v.11 no.4
• /
• pp.57-66
• /
• 2011

At this time, digital cameras are used in art and forensic science. However, the digital camera has some limitations which need to understand of photograph and lighting. It is a useful paper to make an infrared flatbed scanner. The following processes offer an infrared flatbed scanner development. First, the infrared flatbed scanner changes visible fluorescent lamp to infrared LED. Second, it equips a long-pass filter, which is available to pass over 810nm wavelength, on the glass to complete the optimal infrared flatbed scanner. In addition, it must copy from digital camera to computer directly. The infrared digital camera has disadvantage to always irradiate infrared lamp. Because of difference between visible length and infrared length characteristic, they have different focal distance. This devised scanner for solving mentioned problems does not need to irradiate infrared lamp, and there is not a problem about focal point because the depth of field of flatbed scanner is minimum 2mm. Lastly, the infrared flatbed scanner can make high resolution which is 12,800dpi unlike digital camera. Accordingly, the infrared scanner looks forward to be used in many field of study.

• The Journal of the Korea Contents Association
• /
• v.11 no.4
• /
• pp.11-17
• /
• 2011

This paper aims to realize the information of a subject existing in a still image with objective values. To attain the goal, this research takes the vanishing point that a 2D still image has as the basis and recomposes the still image into a 3D image using a 3D program. Also, in order to set up the axis of the camera necessary to recompose a 3D image, this paper used the lens angle of view that the image has and floor patterns as well. The 3D image completed in this way can measure the size and distance of all subjects in the floor patterns if the size value of a particular reference subject is known, and through this, it can be possible to acquire basic information of a subject that can be either a criminal or a clue in the images of CCTVs or some criminal scene.

• The Journal of the Korea Contents Association
• /
• v.10 no.7
• /
• pp.40-48
• /
• 2010

Reflected-Ultraviolet photography is applied to the evidence of crime, Archaeology, and Dermatology. In the past, Ultraviolet photography was done with standard black-and-white film. Because emulsion of film is more sensitive to near UV light than CCD(Charge Coupled Device) or CMOS(Complementary Metal-Oxide-Semiconductor)of digital camera. In this research, we purpose to improve qulity of ultraviolet photographs and to find out the best alternative digital camera by utilizing a cunsumer digital camera. To achieve theses, we removed IR cutoff filter from digital camera. And by using modified UV pass filter, we verified the increase of image resolution of digital ultraviolet photographs. Also, we analyze reproducibility of digital ultraviolet photographs according to type, size, pixel of image sensor. Furthermore, this research resulted in the development of an practical digital camera system by utilizing a cunsumer digital camera. Eventually, it will contribute to practical use in the various field of digital ultraviolet photographs